An Evolutionary Perspective on Updating Risk and Ambiguity Preferences

Abstract

Using an approach based on evolution and adaptation, we provide foundations for a model of choice under uncertainty based on adaptive preferences, and we show that updating of those preferences in response to new information will respect dynamic consistency even at the cost of violating consequentialism. We argue that our approach can be applied in most contexts with ambiguity, and we establish that adaptive preferences nest variants of many models in the literature on risk and ambiguity as special cases. Since models of either ambiguity aversion or non-expected utility for risk cannot satisfy both dynamic consistency and consequentialism without strong assumptions on beliefs and information structures, our results add a novel perspective to the discourse about which of these properties should be maintained in applications. In addition, a central feature of adaptive preferences and the models they nest is a tight connection between ambiguity attitudes and violations of expected utility in the context of risk.